International Journal of Computer Science Trends and Technology (IJCST) Volume 5 Issue 4, Jul Aug 2017

RESEARCH ARTICLE OPEN ACCESS

Semantic Analysis and Report Generation for Social Content

Ms. Sheela A Badagi [1], Mrs. Shanta Kallur [2]
Department of Computer Science & Engineering
K.L.E.Institute of Technology, Hubballi-30
Karnataka - India.

ABSTRACT
With rapid development of e-commerce the merchants selling the products, later ask their opinions. Most customers express
their opinions on various kinds of entities like products/ services. Extracting the useful insights from internet based application
like facebook helps to create and exchange of user generated content to communicate with others. The reviews have rich source
of information for decision making and sentimental analysis. Sentimental analysis refers to a classification problem where the
main focus is to predict the polarity of words and classify them into positive, negative and neutral feelings, with the aim of
identifying attitude and opinions. In this paper, we develop a web application which will derive social content for a company,
service or a product to take business decisions on semantic analysis.
Keywords: Semantic analysis, Opinion mining, Sentimental classification, Polarity, Social Media.

I. INTRODUCTION interested data is of a big task. The social media data is

Semantic analysis is the study of language, it is the
process of relating syntactic structure. The idea of
transformational generative grammar is introduced by the
syntactic structures. This method uses some of the phrase
rules which can break down the sentences into more then one
parts, and its level of clauses, sentences and paragraphs to the
level of writing as whole to their language independent
language. Social media or social content is computer mediated
technologies that helps to creation and sharing of information,
ideas and other forms of expression via networks.
Now a days, More and more products are sold on the web
due to rapid expansion of e-commerce. So it has become
common approach for merchants to enable their customers to
review and give suggestions (opinions) on the products to
which they are purchased [1]. If each product have a more
number of comments then it may difficult for manufacturer to
maintain the information ,so for that mining of the interested
product features, and identifying the opinion of the sentence
and display the result as positive, negative or neutral
comments.
A. Related Work
Social media uses some of the concepts like computer
science, data mining, machine learning, natural language
processing and algorithms. Social media is the internet based
application that helps to create and exchange the user
generated content. It is easy for the users to communicate with
each other. Some of the challenges to data mining are vast,
noisy, unstructured and distributed type of data. Because in
data mining the data is in the form of unstructured, noisy,
distributed, it is not in a exact format which, the users are
interested. Now a days data is everywhere, mining of
everywhere, overloaded in two forms,
Information Overload
Interaction Overload
Information overload includes photos, blogs. Interaction
overload includes the comments, followers, posts. Data
mining from the social media can help to get millions of users
opinions and behaviours for product, i.e
Product sentiments
Human behaviour
With respect to particular product, the product contains
the positive feedback from the users or negative feedback
from the users else neutral feedback. Based on the reviews
developers are going to improve their business. To enhance
their business our project provides Semantic analysis and
report generation for social content i.e
Extract the data from social media (facebook), select
the interested fields (posts, comments, likes),
interpret the comments for individual page that is
positive or negative or neutral, predict aggregation of
comments and represent them by using pai chart.
Social media like facebook analyzer extract the facebook
data by using the facebook graph API into web facebook
entities ,i.e posts, likes, sentiments of comments.
B. Objectives
Extraction of page data from social media like
facebook by using Graph API.
Tell the presence of the Facebook pages.
Number of posts count, fans count, comments for a
selected page.

ISSN: 2347-8578 www.ijcstjournal.org Page 126

 International Journal of Computer Science Trends and Technology (IJCST) Volume 5 Issue 4, Jul Aug 2017
Number of comments count, shares count, likes Client
count for a selected post. Application Server
Classification of extracted data based on keywords. Social media server to access facebook.
Extracting useful insights as sentimental analysis.
Sentimental analysis of individual comment on page, It involves some of the steps as follows,
that is positive, negative or neutral.
The client requests to application server to get page
Sentimental analysis of aggregation of comments on information.
a page. The application server requests for facebook server.
Facebook server responses for requested info to
C. Aim Of The Proposed System Application server.
Client gets the requested information from
Application server.
Facebook
Extract the Web facebook
Facebook data Analyzer Entities The proposed system can help various business
applications such as e-commerce application to get the money
Fig. 1 Facebook Extracter and to save the money.

Figure 1 shows Facebook extracter. The facebook D. Graph API

analyzer tool is used to extract the facebook page information Facebook graph API can help to generate access token,
by using the graph API. by using this url,

Extract data from the social media site https://developers.facebook.com/tools/explorer/1456

Facebook. 34995501895/
Analyzing the extracted data with TF-IDF
vectorization and multinominal naive bayers E. Generation Of Access Token
algorithm. To select access token within graph API explorer, select
Presentation : get the users interacted get user access token, user permissions like,
information such as; number of posts ,fans User name
count, comments for selected post, comments, User location
shares, total number of likes per post , User posts
prediction of comment to tell positive, negative User likes etc..
or neutral.
F. Access Token Information
II. DATA EXTRACTION PROCESS Access token gives information about, Application name.
This process extracts the Data from the facebook by User of that application.
using graph API. Facebook graph API explorer is used to Test, Validation of that application interms of true and
create, and authenticate API calls and debug responses [4] . false.
Expiration time of that application in Indian standard
format.
(Social Response()
Scopes tell about the fields which are accessible by
Media For page
the user.
Server)
Internet
III. SENTIMENT ANALYSIS
orRequest
Cloud () for page
Sentiment analysis (Opinion analysis) can be achieved
with the help of Sentiment classification [5]. Emotions are the
Request () for page positive or negative reactions. Human communication is under
the control of emotions and opinions [10]. While
communicating, humans can express their opinions by face to
Client Response () for page face. But now a days, e-commerce growing faster, and all are
using internet to share the information [6].
Client App Server Sentiments can be classified based on,
Fig. 2 Facebook analyzer Emotions
serv
Figure 2 shows Facebook analyzer. The proposed system tells Keywords
about over all objectives of this paper, system designercontains Sentiwordnet
3 components as follows,

ISSN: 2347-8578 www.ijcstjournal.org Page 127

 International Journal of Computer Science Trends and Technology (IJCST) Volume 5 Issue 4, Jul Aug 2017
In this paper, sentiments can be classified with the help of
positive and negative keywords. Positive keywords are like,
gud, super, excellent, efficient and so on. Negative keywords
like bad, worst, cheat, not good, etc... by using the
multinomial naive bayers algorithm we can predict the
polarity of the keywords is positive or negative. In other cases,
the same keywords give the different meaning in different
situation.
Example: Battery life of the Samsung cell is too long (positive
comment).
Samsung mobile takes too long to start (negative comment).
G. Natural Language Processing(NLP)
It is the field of interaction between computer and human
(natural) language [7].
NLP contains
NLG(Natural language generation)
NLU(Natural language understanding)
The natural language generation (NLG) converts computer
database language to human readable form. The natural
language understanding (NGU)is the identification of single
semantic from multiple semantics by using NLP.
H. Uses Of NLP Algorithms
NLP algorithm based on machine learning algorithms,
how the data can be analyzed more is the model accuracy.
NLP algorithms are useful to create the advanced applications
and prototypes.
Summarize blocks of text: Ignoring irrelevant
information, only extracting the useful information.
Identify the type of entity extracted: Identify the
named entity recognition.
Sentimental analysis: Identify the sentiment of a
string from text, that is positive, negative or neutral.
With the help of Tf-Idf vectorizer and naive bayers algorithm
we are going to classify the text within document.
K. Tf-Idf Vectorizer
TfIdf can be successfully used for stop-words filtering in
various subject fields including text summarization and
classification.
Tf(Term Frequency) tells about the number of times
a term occurs in a document is called its term
frequency.
Idf(inverse Document Frequency)tells about weight
of terms that occur very frequently in the
document[8].
Equation (4) gives the Term frequency weight is,
tf(t,d)=f t,d .............................................................(4)
Equation (5) gives the Inverse document frequency weight is,
.......................(5)
Where.
N-Total number of documents.
{ }-Number of documents where term appears.
IV. DISCUSSION AND RESULTS
Facebook analyzer is a web application. It searches and
analyzes the facebook pages for entered keyword. If pages
available then it displays all related pages. It also shows posts,
fans count for that page. And for each post it displays number
of comments, number of shares and count of likes.
Eg: Go to browser and enter the url as: localhost 5000.
Keyword entered is: Narendra Modi

I. NLP Examples
Sentimental analysis: Based on stand ford (tool)
NLP can help to identify the opinion (positive,
negative, neutral [9]).
Social media analysis: Extract subjective
information about particular object based on polarity
of object, identify the trends of public opinion.

J. Multinomial Naive Bayers Algorithm

Naive bayers algorithm is used for text classification.
For a document d and a class c, equation(1) gives the
conditional probability is,

P(c/d)=P(d/c)p(c)/p(d) .....................(1)
Fig. 3 Show Pages for entered keyword.
Cmap=argmax CC P(c/d)...................(2)
The Fig 3 shows the facebook analyzer, that searches the
All word positions in test document can be consider, pages for entered keyword by using, any of the browsers.
CNB=argmaxcjcP(cj)ipositions P(xi/cj)
.....................................................................(3)

ISSN: 2347-8578 www.ijcstjournal.org Page 128

 International Journal of Computer Science Trends and Technology (IJCST) Volume 5 Issue 4, Jul Aug 2017
The above figure 6 gives the number of comments,
shares, likes, along with, all comments can be displayed.
Prediction of these individual comments helps to provide the
sentimental analysis. Aggregation analysis of these comments
can be reported by using chart.

Fig. 4 Related pages for entered keyword.

The Fig 4 shows all related pages for given keyword

Narendra modi. i.e, Narendra modi, I support Narendra modi, Fig.7 Prediction of single comment.
Narendra modi fans from karunaadu so on.
The above figure 7 shows the, Each comment can be
predicted as positive, negative and neutral as shown above.

V. CONCLUSION
This project helps to analyze various user requests as
positive, negative, neutral. It even aids, getting aggregation of
comments, which may help to take a decision for a given
query. This System provides system owner or user to predict
upon future plan of trends.

REFERENCES

Fig. 5 Show posts, post counts, fans counts.

[1 [1] Minqing Hu and Bing Liu Department of Computer
Science University of Illinois at Chicago 851 South
The above figure 5 gives the post count, fans count for Morgan Street Chicago, IL 60607-7053 {mhu1,
that selected page. each post contains post id, created time and liub}@cs.uic.edu
messages. The messages may be in all languages. But it can [2]
predict only English comments. https://en.wikipedia.org/wiki/Google_Analytics#Populari
ty
[3] https://en.wikipedia.org/wiki/Social_media_analytics
[4] https://developers.facebook.com/tools-and-support/
[5 International Journal of Innovative Research in
Advanced Engineering (IJIRAE) ISSN: 2349-2163
Volume 1 Issue 10 (November 2014)
[6] International Journal of Computer Applications
Technology and Research Volume 4 Issue 6, 419 - 424,
2015, ISSN:- 23198656
[7]
https://en.wikipedia.org/wiki/Natural_language_process
ing.
[8] https://en.wikipedia.org/wiki/Tf%E2%80%93idf.

[ [9] International Journal of Advanced Research in Computer

Fig. 6 Analysis of comments for a single post. Science and Software Engineering (Volume 5, Issue 5,
May 2015)

ISSN: 2347-8578 www.ijcstjournal.org Page 129

International Journal of Computer Science Trends and Technology (IJCST) Volume 5 Issue 4, Jul Aug 2017
[10] Information Sciences Institute University of Southern
California 4676Admiralty Way Marina del Rey, CA
90292-6695 skim@isi.edu Information Sciences
Institute University of Southern California 4676
Admiralty Way Marina del Rey, CA 90292-6695
hovy@isi.edu Proceedings of the COLING
conference, Geneva, 2004

ISSN: 2347-8578 www.ijcstjournal.org Page 130